cornzz

Exception when opening plugin interface and update to JUCE 5.4.6

1

The exception is only thrown in debug mode since it is triggered by a valid coordinates assertion during `currentPositionMarker.setRectangle()` in `AudioThumbnailComp.cpp`, caused by `timeToX()` returning -nan(ind) and therefore the new...

Support for Solr 9?

3

Hi there, are you planning on supporting Solr version 9?

I would've loved to use this simple package for our purposes but one of your dependencies, `soupselect`, hasn't been updated since 13 years and therefore itself has some dependencies with...

[Question]: LLMLingua-2 Sample-Wise Dynamic Compression Ratio

2

### Describe the issue Hi, I have two questions: 1. Appendix L of the LLMLingua-2 paper talks about allowing the compressor to adjust the compression rate for different samples, but...

Playground broken in firefox

2

### Describe the bug The [gradio playground](https://www.gradio.app/playground) seems to be broken in firefox, the preview window stays blank and the following error shows up in the console: ``` Uncaught (in...

### Describe the bug This is not a big problem, I just noticed that my gradio app, which runs behind a reverse proxy on the path `/app`, is throwing this...

### Describe the issue 1. The publicly available validation split only contains a small number of samples (around 20 per task), and the ground truths are not available for the...

[Bug]: LLMLingua-2 uses wrong special tokens by default

2

### Describe the bug The `TokenClfDataset` is [initialized without a `model_name`](https://github.com/microsoft/LLMLingua/blob/60abc0f94939b24e000fe6a33a954de72055fa0c/llmlingua/prompt_compressor.py#L2341) parameter and therefore [defaults to `bert-base-multilingual-cased`](https://github.com/microsoft/LLMLingua/blob/60abc0f94939b24e000fe6a33a954de72055fa0c/llmlingua/utils.py#L19), meaning that incorrect special tokens are used in llmlingua-2, i.e. ```python if "bert-base-multilingual-cased"...

Fix wrong special tokens being used for llmlingua-2

1

# What does this PR do? Fixes #181 The `model_name` parameter was not set when initializing `TokenClfDataset` which lead to the special tokens (bos/eos/pad) of the llmlingua-2-small model being used...

Fixes https://github.com/mistralai/mistral-inference/issues/215 Attention bias was being created on cuda:0 regardless of the selected cuda device as the correct device was not being passed to `from_seqlens()` in `BufferCache.get_input_metadata()`